Fail to run the notebook

[fzhu2e]

Hi, first thanks for the nice work on this topic.
I was trying to run through the notebook example_Su_etal2015.ipynb but got the error below.
It seems that the MATLAB session was not returning the correct type.
I know it has been years, but do you have an idea how to fix this? Thanks!

Starting MATLAB on ZMQ socket ipc:///tmp/pymatbridge-662a070a-85d8-4f2e-a081-7c3e6ac8db61
Send 'exit' command to kill the server
............MATLAB started and connected!
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-2-82f10c014ec8> in <module>()
      1 from hhtpywrapper.eemd import EEMD
      2 
----> 3 eemd_post_processing = EEMD(rate, 6.0, 100, num_imf=10, seed_no=4, post_processing=True)
      4 eemd_post_processing.get_oi()

~/.pyenv/versions/anaconda3-5.0.1/envs/py3.6/lib/python3.6/site-packages/hhtpywrapper-0.1.dev0-py3.6.egg/hhtpywrapper/eemd.py in __init__(self, signal, Nstd, NE, num_imf, run_CEEMD, max_sift, type_spline, modify_BC, rand_type, seed_no, check_signal, post_processing)
     12                             run_CEEMD, max_sift, type_spline, modify_BC,
     13                             rand_type, seed_no, check_signal)
---> 14         allmode = res['result'].T
     15         num_imf = allmode.shape[1]
     16         if post_processing == True:

AttributeError: 'str' object has no attribute 'T'

[YihaoSu]

Hi, I just tested the code and it works fine (see the screenshot below). It seems that the res['result'] returns "some message" (str type) instead of calculation value when you executed the code. Did you change the input file? If you use different input signal, you have to change parameters of the EEMD class (e.g., Nstd: noise level, NE: ensemble number) to get physically meaningful intrinsic mode functions (IMFs).
I have tried numerous parameter sets and this parameter set works my data. But the parameter set is not suitable for different input signal.

[fzhu2e]

Thanks for the quick response! I didn't change the input. Oh, but I am on macOS. Is that the cause of this? Is there a slow EEMD version that I can use on macOS?

[YihaoSu]

Hm...
If you restart the Jupyter notebook kernel, would it help?
I did encounter this situation when I run large data set (memory not enough) or pymatbridge automatically close session for no reason.

You can first check the rate variable to see if it input correctly or test small dataset (rate=rate[:10000]). You may also hardcode the eemd.py to print "some message" of res['result'].

BTW, it takes me about 1 min to finish the fast EEMD of my QPO dataset on my Liunx OS. The file eemd.m in the HHT MATLAB program package
implements the old, slow version of EEMD algorithm. In my experience, it is at least 10 times slower than the fast EEMD code.